Temporal and Regional Analysis of Earthquake Activity in Turkey (1975–2023)

Welcome to our project page.

Here, you’ll find updates and insights from our ongoing exploration of earthquake patterns in Turkey.

1. Project Overview and Scope

Earthquakes are one of the most important natural hazards in Turkey, a country that located on several active fault lines — especially the North Anatolian Fault Zone. Understanding when and where earthquakes happen is essential for planning, risk reduction, and disaster preparation.

This project analyzes seismic activity in Turkey from 1975 to 2023, using a dataset published by the United States Geological Survey (USGS). The dataset includes thousands of events with details such as time, location, magnitude, and depth, all limited to Turkey’s geographical borders.

The main question of this research is:

“How did earthquake frequency and magnitude change in Turkey over time, and which regions show the most consistent seismic activity?”

To answer this question, the project sets the following objectives:

  • Temporal Analysis: Track annual or seasonal shifts in earthquake frequency and intensity.
  • Spatial Mapping: Visualize the geographic clustering of earthquakes to identify high-risk zones.
  • Magnitude Patterns: Investigate the distribution of magnitudes and how often stronger quakes occur.
  • Risk Identification: Identify areas of Turkey that exhibit consistently high seismic activity and may require closer monitoring or mitigation efforts.
  • Visualization: Produce clear and informative visualizations, including histograms, time series plots, and density maps, to support interpretations.

All analyses are performed in R, and the project is compiled and shared through a Quarto-powered GitHub page, making it fully transparent and reproducible. The final product is intended to be both informative and accessible to audiences interested in Turkey’s seismic landscape—whether in research, policy, or planning.

2. Data

Understanding the nature of seismic activity in Turkey requires a data-driven approach built on reliable, long-term records. This project builds on a dataset that spans almost three decades, offering a broad view of how seismic behavior has unfolded in different parts of Turkey. It allows not only for time-based comparisons, but also for mapping out regions with higher seismic exposure—making it an ideal foundation for exploratory and risk-focused analysis.

2.1 Data Source

The dataset used in this project was obtained from the USGS Earthquake Catalog, a widely used open-access platform that provides detailed global seismic activity records. For this project, we filtered the data to include only earthquakes that occurred within the borders of Turkey — specifically between 26°–45° Eastern longitudes and 36°–42° Northern latitudes — between the years 1975 and 2023. Additionally, only events with a magnitude of 2.0 or greater were included to focus on perceptible and potentially damaging seismic activity.

The dataset was downloaded in .csv format and contains information such as:

  • Date and Time of the earthquake
  • Magnitude
  • Depth (in kilometers)
  • Latitude and Longitude
  • Place (a descriptive region name)

After downloading, the data was imported into R for further exploration and preprocessing.

2.2 General Information About Data

The dataset consists of approximately 30,000 earthquake events recorded in Turkey over a 48-year period. Each row in the dataset corresponds to a unique seismic event, with detailed information such as the time it occurred, its geographical coordinates, how deep it originated, and how strong it was. Most earthquakes are naturally concentrated around Turkey’s well-known fault zones, such as the North Anatolian Fault in the north and the East Anatolian Fault in the southeast. This spatial clustering is something we aim to explore visually in the later sections of the project.

Here’s a brief overview of the variables:

  • time: Timestamp of the earthquake (UTC), including date and hour
  • latitude & longitude: Geographic location of the event
  • depth: Depth below the surface in kilometers
  • mag: Magnitude on the Richter scale
  • place: Descriptive label provided by USGS, such as nearby city or region

Overall, The structure of the dataset is tidy and consistent, which makes it easier to analyze without extensive restructuring. Thanks to its clean structure, the dataset served as an effective and convenient entry point for our analysis.

2.3 Reason of Choice

There are several reasons why we chose this specific dataset:

  • First and foremost, earthquakes have always been a critical topic in Turkey, especially following the devastating seismic events in 1999 and more recently in 2023. Understanding patterns in historical data can offer valuable insights into risk mitigation.
  • Second, the dataset is both rich and clean, and doesn’t require extensive preprocessing to begin analysis. It offers a long temporal range (1975–2023), which makes it ideal for trend detection.
  • Lastly, We wanted a dataset that is directly relevant to public safety and urban planning in Turkey. The output of this study could contribute to further academic or policy-oriented discussions about seismic preparedness.

2.4 Preprocessing

After importing the CSV file into R, several preprocessing steps were performed to prepare the dataset for analysis:

  • Datetime Conversion: The time column was converted to proper date-time format to make time-based analysis possible.
  • Filtering: Earthquakes smaller than magnitude 2.0 were left out, as they’re too weak to be relevant for a risk-focused study.
  • Region Extraction: The place was used to pull out province-level location info when available.
  • Derived Variables: A year column was created from the datetime to allow annual comparisons.
  • Missing Values: The dataset had very few missing values, which were either removed or reasonably filled in.

The cleaned dataset was saved in .RData format for reproducibility and faster processing in future sessions.

Show the code
eq_data <- read.csv("turkey_earthquake_data.csv")
eq_data$DateTime <- as.POSIXct(eq_data$time, format="%Y-%m-%dT%H:%M:%OSZ", tz="UTC")
eq_data <- subset(eq_data, mag >= 3.0)
eq_data$year <- format(eq_data$DateTime, "%Y")
eq_data <- eq_data[ , !(names(eq_data) %in% c("magType", "nst", "gap", "dmin", "rms", 
                                              "net", "id", "updated", "type", 
                                              "horizontalError", "depthError", 
                                              "magError", "magNst", "status"))]
save(eq_data, file="eq_data.RData")

Sample of the Cleaned Dataset

The table below presents the first few rows of the cleaned earthquake dataset used in this study. After filtering out earthquakes below magnitude 3.0 and removing irrelevant columns, the dataset now includes key attributes such as date, location, magnitude, and depth. This refined structure allows for a more focused analysis of significant seismic activity in Turkey from 1975 to 2023.

Show the code
library(knitr)

# İlk 10 satırı daha düzgün biçimli tablo olarak göster
kable(head(eq_data, 10), caption = "Table 1: Sample of Pre-processed Earthquake Data")
Table 1: Sample of Pre-processed Earthquake Data
time latitude longitude depth mag place locationSource magSource DateTime year
1975-04-23T01:08:08.500Z 40.450 26.076 20 4.3 30 km S of Enez, Turkey us us 1975-04-23 01:08:08 1975
1975-04-30T04:28:56.900Z 36.184 30.772 56 5.6 41 km SSE of Tekirova, Turkey us us 1975-04-30 04:28:56 1975
1975-05-02T05:03:19.300Z 36.221 30.612 43 3.9 30 km ESE of Beykonak, Turkey us us 1975-05-02 05:03:19 1975
1975-05-07T17:59:15.800Z 40.370 26.092 33 3.9 24 km NE of Gökçeada, Turkey us ath 1975-05-07 17:59:15 1975
1975-05-30T05:13:45.100Z 39.118 27.747 10 3.9 6 km E of K?rka?aç, Turkey us us 1975-05-30 05:13:45 1975
1975-05-30T14:22:40.800Z 38.732 27.620 9 4.2 4 km E of Saruhanl?, Turkey us ath 1975-05-30 14:22:40 1975
1975-05-31T05:36:52.900Z 36.162 29.015 42 4.0 37 km WSW of Kalkan, Turkey us us 1975-05-31 05:36:52 1975
1975-05-31T12:41:23.900Z 36.642 28.131 19 4.0 19 km SSW of ?çmeler, Turkey us us 1975-05-31 12:41:23 1975
1975-06-02T03:19:07.200Z 36.652 26.590 25 4.5 24 km ENE of Astypálaia, Greece us us 1975-06-02 03:19:07 1975
1975-06-04T02:55:37.000Z 40.857 31.513 68 3.8 15 km NNW of Bolu, Turkey us us 1975-06-04 02:55:37 1975

3. Analysis

In this part of the project, we took a closer look at Turkey’s earthquake data to spot trends and patterns in seismic activity. We started by getting familiar with the dataset—checking its overall structure, looking at the spread of magnitudes and depths, and making sure the data was clean enough to work with. There were a few missing values, but most of them wasn’t a big issue and were handled early on.

Then, we grouped the earthquakes by year to see how their frequency and intensity changed over the 28-year period. This helped us track how things evolved over time. At the same time, we also mapped out where earthquakes were happening across the country, which helped us find hotspots and regions with repeated activity.

We tried to see if any variables were related—for example, we looked at whether deeper earthquakes tend to be stronger, though the results weren’t always clear. Later on, we used some basic clustering methods and heatmaps to highlight areas with more intense seismic behavior. These visuals made it easier to figure out which regions might need more attention or monitoring.

All the analysis was done in R, using tools that are great for working with spatial data and making visuals. Overall, the findings give a useful starting point for deeper studies and could help people working in planning, safety, or research to better understand Turkey’s earthquake patterns.

3.1 Exploratory Data Analysis

In this section, we will conduct an exploratory analysis of the earthquake data to understand its basic structure and identify potential patterns or issues.

3.1.1 Overview of the Dataset

The dataset includes information on earthquakes that occurred in Turkey from 1975 to 2023, filtered to include only events with a magnitude of 3.0 or higher. After preprocessing, the dataset consists of the following key columns:

  • DateTime: The exact date and time of each earthquake (in UTC).
  • latitude: The north–south coordinate of the event.
  • longitude: The east–west coordinate of the event.
  • depth: The depth of the earthquake’s focus, in kilometers.
  • mag: The magnitude of the earthquake, based on the Richter scale.
  • place: A textual description of the location, such as nearby towns or provinces.
  • year: Extracted from the date, useful for time-series grouping.

The dataset is structured in a tidy format with no missing values in the key analytical fields.

Show the code
sum(is.na(eq_data))
[1] 0

3.2 Visualization and Analysis

Understanding the distributions of key variables gives insight into the data’s characteristics and helps identify any anomalies or patterns.

Earthquake Magnitude Distribution

The magnitude of earthquakes is fundamental to assessing seismic risk. This histogram shows how often different magnitude values occur in the dataset.

Show the code
library(ggplot2)
ggplot(eq_data, aes(x = mag)) +
  geom_histogram(binwidth = 0.2, fill = "#7686A3", color = "black") +
  labs(title = "Distribution of Earthquake Magnitudes",
       x = "Magnitude", y = "Frequency")

Earthquake Depth Distribution

Depth indicates how deep beneath the surface an earthquake originates. This histogram helps us understand the typical depth range of earthquakes in Turkey.

Show the code
ggplot(eq_data, aes(x = depth)) +
  geom_histogram(binwidth = 5, fill = "#99BAB6", color = "black") +
  labs(title = "Distribution of Earthquake Depths",
       x = "Depth (km)", y = "Frequency")

Earthquake Frequency Over Time

Next, we analyze how earthquake occurrences vary over the years. This temporal view is crucial for detecting trends or changes in seismic activity.

Show the code
eq_data$year <- format(as.Date(eq_data$DateTime), "%Y")
earthquakes_per_year <- table(eq_data$year)

ggplot(data.frame(year = names(earthquakes_per_year), count = as.numeric(earthquakes_per_year)),
       aes(x = year, y = count)) +
  geom_bar(stat = "identity", fill = "#27394A") +
  theme(axis.text.x = element_text(angle = 90)) +
  labs(title = "Earthquake Frequency by Year",
       x = "Year", y = "Count")

Spatial Distribution of Earthquakes

Mapping earthquake epicenters provides spatial insights and helps identify regions of concentrated seismic activity.

Show the code
library(ggplot2)
library(sf)
Linking to GEOS 3.13.0, GDAL 3.8.5, PROJ 9.5.1; sf_use_s2() is TRUE
Show the code
turkey_shapefile <- st_read("/Users/akbulut/Documents/GitHub/emu660-spring2025-edagrlk/turkey.shp")
Reading layer `turkey' from data source 
  `/Users/akbulut/Documents/GitHub/emu660-spring2025-edagrlk/turkey.shp' 
  using driver `ESRI Shapefile'
Simple feature collection with 250 features and 168 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 83.6341
CRS:           NA
Show the code
ggplot() +
  geom_sf(data = turkey_shapefile, fill = "white", color = "black") +
  geom_point(data = eq_data, aes(x = longitude, y = latitude, color = mag), alpha = 0.5) +
  scale_color_viridis_c(option = "inferno", direction = -1) +
  coord_sf(xlim = c(26, 45), ylim = c(36, 42), expand = FALSE) +
  labs(title = "Spatial Distribution of Earthquakes in Turkey",
       x = "Longitude", y = "Latitude", color = "Magnitude") +
  theme_minimal()      

3.2.1. Spatial Distribution of Earthquakes in Turkey: Top 10 Years by Frequency

This series of maps shows where earthquakes occurred during the ten years when Turkey experienced the highest levels of seismic activity. By focusing on these especially active years, we can get a better sense of how earthquake epicenters were spread out across different parts of the country during times when the ground was particularly restless.

On each map, earthquakes are colored based on their magnitude, which makes it easier to pick out the stronger ones at a glance. To make sure significant events aren’t overlooked, we also displayed earthquakes over magnitude 5 as larger points that stand out more clearly.

It’s important to keep in mind that the high number of earthquakes in some of these years might be partly due to swarms of aftershocks following major events. These aftershocks, while smaller, still contribute to the overall seismic activity and can create noticeable clusters in certain regions.

These maps help us spot patterns—like which areas tend to shake more during active periods. Recognizing these hotspots is a key step for understanding risk and making smarter decisions about safety and planning in those regions.

Show the code
library(ggplot2)
library(sf)

turkey_shapefile <- st_read("/Users/akbulut/Documents/GitHub/emu660-spring2025-edagrlk/turkey.shp")
Reading layer `turkey' from data source 
  `/Users/akbulut/Documents/GitHub/emu660-spring2025-edagrlk/turkey.shp' 
  using driver `ESRI Shapefile'
Simple feature collection with 250 features and 168 fields
Geometry type: MULTIPOLYGON
Dimension:     XY
Bounding box:  xmin: -180 ymin: -90 xmax: 180 ymax: 83.6341
CRS:           NA
Show the code
top10_years <- names(sort(table(eq_data$year), decreasing = TRUE))[1:10]

for (yr in top10_years) {
  eq_year <- subset(eq_data, year == yr)
  
  cat("\n\n### Earthquake Distribution for Year:", yr, "\n\n")
  
  print(
    ggplot() +
      geom_sf(data = turkey_shapefile, fill = "white", color = "black") +
      geom_point(data = eq_year,
                 aes(x = longitude, y = latitude, color = mag),
                 alpha = 0.6,
                 size = ifelse(eq_year$mag > 5, 5, 2)) +  # 5'ten büyükse 5, değilse 2
      scale_color_viridis_c(option = "inferno", direction = -1) +
      coord_sf(xlim = c(26, 45), ylim = c(36, 42), expand = FALSE) +
      labs(title = paste("Spatial Distribution of Earthquakes in Turkey - Year:", yr),
           x = "Longitude", y = "Latitude", color = "Magnitude") +
      theme_minimal()
  )
}


### Earthquake Distribution for Year: 2005 



### Earthquake Distribution for Year: 1994 



### Earthquake Distribution for Year: 1999 



### Earthquake Distribution for Year: 2023 



### Earthquake Distribution for Year: 2004 



### Earthquake Distribution for Year: 1993 



### Earthquake Distribution for Year: 2007 



### Earthquake Distribution for Year: 2006 



### Earthquake Distribution for Year: 1991 



### Earthquake Distribution for Year: 1995 

The earthquake dataset used in this study reliably captures the spatial and temporal distribution of seismic events across Turkey, including some of the country’s most significant and devastating earthquakes. Notably, large earthquakes such as the 1999 İzmit earthquake, the 2023 Kahramanmaraş earthquake, the 1995 Dinar earthquake, and the 2005 Bingöl earthquake are clearly represented with corresponding high-magnitude events and dense epicenter clusters in the respective years.

This realistic portrayal strengthens the credibility of the data, as it aligns well with well-documented seismic history and major catastrophe timelines. Such consistency supports the use of this dataset for both analytical insights and risk assessment applications.

Regarding human and economic impacts, the following major earthquakes have been associated with severe consequences in terms of fatalities and structural damage:

Year Earthquake Estimated Death Toll Estimated Damage (in USD) Source
1991 Erzincan Earthquake ~750 Severe USGS
1993 Erzincan Earthquake ~1500 Severe EM-DAT
1994 Marmara (Ladik) Earthquake Minimal
1995 Dinar Earthquake ~90 Significant Kandilli Observatory
1999 İzmit Earthquake ~17,000 Estimated billions AFAD, USGS
2004 Bolu-Gerede Earthquake ~150 Considerable EM-DAT
2005 Bingöl Earthquake ~176 Major USGS
2006 Düzce Earthquake ~845 Major AFAD
2007 Balıkesir Earthquake Minimal Moderate USGS
2023 Kahramanmaraş Earthquake >50,000 Massive ReliefWeb, AFAD

Note: The death tolls and damage estimates vary between sources but consistently highlight the catastrophic nature of these events. Note: In 1994, an earthquake occurred in the Marmara region of Turkey. Although it was a noticeable seismic event, it did not result in significant destruction or a high number of casualties compared to other major earthquakes in the country. Therefore, its impact is considered minor relative to catastrophic events such as those in 1999 and 2023.

By comparing our earthquake data with well-known seismic events in Turkey, we were able to confirm that the dataset accurately captures both the timing and strength of the country’s major earthquakes. This allows us to confidently use it as a basis for spatial and temporal analyses as well as for informing risk mitigation strategies.

4. Results and Key Takeaways

When we looked at earthquakes in Turkey between 1975 and 2023, we wanted to understand not just how often they happened but also where they were most intense. By combining the time and place of these events, we uncovered patterns that show which years were particularly active and which regions faced the most powerful shakes. These findings help us get a clearer picture of earthquake risks in Turkey and can guide efforts to better prepare and protect communities.

Here are the main takeaways from our work:

• This study takes a close look at the timing and locations of earthquakes in Turkey over nearly three decades, using detailed data from the USGS earthquake catalog. • We found that certain years stood out with more earthquake activity, and we mapped these events to highlight hotspots with frequent and strong quakes. • The dataset reflects real-life major earthquakes, like the devastating ones in 1999 and 2023, making our analysis grounded in actual seismic history. • By spotting clusters of big earthquakes, we identified areas that might need extra attention when it comes to risk planning and disaster readiness. • Overall, this work shows how looking at both when and where earthquakes happen can give a much better understanding of the risks, helping us take smarter steps to reduce them.

*ChatGPT was utilized in this project.

Download PDF

Back to top